Overview
- Researchers at USC trained a deep‑learning model on about 14,748 research‑quality MRI scans to compute a 'local brain age' value for each voxel, producing anatomically detailed maps instead of a single global brain‑age number.
- When tested on more than 1,900 ADNI scans, the maps showed systematically older‑appearing frontal and temporal regions versus parietal and occipital areas and a slight right‑hemisphere bias.
- People with mild cognitive impairment or Alzheimer’s disease had accelerated local brain age in regions hit early by Alzheimer’s pathology, including the hippocampus, amygdala and several deep nuclei.
- Higher local brain age in these regions correlated with poorer scores on cognitive tests, with the strongest relationships seen in people with Alzheimer’s disease.
- Authors stress this is a research advance that needs validation on more diverse clinical datasets and in longitudinal studies to test whether the maps can predict decline or track treatment effects.